1. Technical Field
The present disclosure relates to video clustering and, more specifically, to clustering crowdsourced videos by line-of-sight.
2. Discussion of Related Art
Mobile crowdsourced videos, which are videos captured and contributed by large groups of individuals, have remarkable value. Uploaded videos often present engaging, diverse, and comprehensive perspectives not captured by professional videographers. Broad appeal of user-uploaded videos has been widely confirmed: freely distributed on YouTube, by subscription on Vimeo, and to peers on Facebook/Google+. Unfortunately, user-generated multimedia can be difficult to organize; these services depend on manual tagging or machine-mineable viewer comments. While manual indexing can be effective for popular, well-documented videos, newer content and content not manually tagged, may be poorly searchable. Live video, or video that is acquired in near real-time is almost impossible to search for.